Comparisons of gyrofluid and gyrokinetic simulations

Abstract

The gyrokinetic and gyrofluid models show the most promise for large scale simulations of tokamak microturbulence.This paper discusses detailed comparisons of these two complementary approaches.Past comparisons with linear theory have been fairly good, therefore the emphasis here is on nonlinear comparisons. Simulations include simple two dimensional slab test cases, turbulent three dimensional slab cases, and toroidal cases, each modeling the nonlinear evolution of the ion temperature gradient instability. There is good agreement in both turbulent and coherent nonlinear slab comparisons in terms of the ion heat flux, both in magnitude and scaling with magnetic shear. However, the nonlinear saturation level for 1¢1 in the slab comparisons show differences of approximately 40%.Preliminary toroidal comparisons show agreement within 50%, in terms of ion heat flux and saturation level.

Nearest-Grid-Point Interpolation in Gyrokinetic Particle-in-Cell Simulation

“…We now describe a very simple two-dimensional slab model without magnetic shear which is a minimal model for nonlinear saturation of the ITG instability [21,22]. Here, we keep the evolution of the purely radial modes, which was not done in earlier simulations of this type [21][22][23].…”

“…Here, we keep the evolution of the purely radial modes, which was not done in earlier simulations of this type [21][22][23]. This model shows the generation and saturation of zonal flow modes in a very simple system.…”

“…The simulation has been linearly benchmarked with the results of Ref. [22]. The most unstable mode is (1, 1), l = 1, m = 1, where k x = l/L x and k y = 2m/L x .…”

“…[21,22] for details on the mode-coupling theory). However, the generation of the purely radial ␦T (2, 0) mode generates a corresponding purely radial electric field [26] (2, 0) ≈ −…”

“…It has been found that without the zonal flow generation, the nonlinear saturation is due primarily to wave-particle trapping, and that the saturation level can be estimated by balancing the growth rate with the trapping frequency [22], producing || = 1 2 ␥ k x k y using the dimensionless units discussed above. This estimate for the saturation level gives 0.63%, which compares well with the 0.61% shown in Fig.…”